I will be adding a couple of additional ESX hosts in the coming months.
If I want to go quadcore so right now my only current options are Intel based boxes. My current hosts are 2 dual core Opteron 2 quad socket boxes (DL585) and 2 dual socket boxes (DL385).
I'm going to be moving the dual socket hosts in to the same HA/DR cluster that the quad sockets servers are in the coming months as I upgrade them to 3.0X. They were implemented by another team and they never upgraded. They have the same CPUs as my quad socket boxes.
Is the lack of an integrated memory controller able to be mitigated by the huge amounts of cache on the newer Xeon chips in a dual socket box,
or is it worth waiting for boxes with quadcore Opterons to become available?
This depends on your VM types. If you have 4 vCPU VMs, then quad cores are very very nice. If you do not really have 4 vCPU VMs. Do you really need quad core? Are you running out of CPU or Memory? THis would dictate if I should wait or not.
If you are running 64bit VMs, then all the Hosts in your clusters should be the same type so that vMotion works. If they are 32bit then you can setup the masks hopefully to do vMotion from AMD to Xeon and back with no issues.
Personally I like a huge amount of cache, it helps alleviate cache misses and that is very important for performance reasons. Specifically if you are running a lot of VMs with the same code base.
The nice thing about going with Dual core boxes now, is that in a year or so when you may need more CPU power you can upgrade to a quad core. It is a thought....
I really was not considering buying dual core given VMWare's current licensing policy. The question is whether to wait for AMD's quad core implementation or goes with Intel's. If I was buying more quad socket boxes there would be no question that I would wait and go with AMD.
It will be interesting to see the power consumption comparison at the wall socket once Barcelona based servers are available. FB Dimms use a lot more power.
The following link is not a suggestion to wait and choose AMD over Intel but it is rather an example of how important things (in my opinion) are overlooked and we sometimes get lost in the details:
More cores also helps with with 2 vSMP and even single vCPU as it gives the scheduler more to work with.
Seems to me upgrading CPUs in enterprise class servers (IBM/HP etc) doesn't work out to be a smart option given the high cost of the replacement CPUs in comparison to an entire new server.
Hey, I hope you folks aren't still waiting! Appears it'll still be a while...
One thing I haven't seen mentioned... will the product be released in only a 2Ghz model, with speed boosts starting a quarter later? That was the plan during the initially scheduled release. AMD may not have product on the shelves until Intel ships its second generation 45-nm Nehalems! This is a huge misstep by AMD.
Hopefully they will be the 2350, 2.5 ghz.
I did see some interesting comment in a review that compares the upcoming Penryn to the 2350s.
"While AMD loses quite a few battles, the war is far from lost. The server/HPC situation is entirely different from the desktop scene where the Core 2 Quad overpowers the Phenom in almost every benchmark. There is more to server and HPC performance than simple raw processing power. Intel's flagship still has an Achilles heel: the platform it is running on has higher latency and much lower bandwidth than AMD's platform. Once you really stress all those cores with many threads, AMD's platform starts to pay off."
"To put it in car terms, our SPECjbb, LINPACK, and MySQL benchmarks have shown that Intel's "powerful CPU engines" sometimes have problems putting the "massive torque" to the "wheels". You may feel for example that using four instances in our SPECjbb test favors AMD too much, but there is no denying that using more virtual machines on fewer physical servers is what is happening in the real world. Intel's best have a solid lead over AMD's quad-core in rendering benchmarks, but some HPC, Java and MySQL benchmarks show that the 2.5GHz Barcelona is able to keep up with (or come close to) a 3GHz Xeon 5472. That is impressive, on the condition that we finally see some higher clocked Opteron 23xx chips in commercially available servers."
Message was edited by: pdrace
Yes, but if you need a solid platform today, with the highest performance, non-existant AMD chips aren't going to get the job done. Once they are finally out, I think all the current benchmarks show that the platforms are pretty close - which is a good thing. Looking at specific workloads, cases can be made for each platform performance-wise.
But in the grand scheme of things, does a 10% advantage for any workload make a particular platform that appealing, when compared to the previous generations (which these blow away)?? Not to me when considering our virtualization infrastructure supports a general workload. My bigger issue is having a platform available that fits within my budget, giving me the best bang for my buck long term.
I don't need the servers immediately so I'm waiting to see how things play out. If AMD doesn't get it's act together in Q1 of 08 I will have to go with Intel quads at some point. I will stick with 2 socket boxes though. 4 CPUs on the FSB makes me nervous. I'm not going to assume that Nehalem is going to be terrific. This will be Intel's first generation of chips with a IMC and no FSB. AMD's IMC and Direct connect architechture is mature.